Working memory deficits are present in many neuropsychiatric diseases with diagnosis-related severity. However, it is unknown whether this common behavioral abnormality is a continuum explained by a neural mechanism shared across diseases or a set of discrete dysfunctions. Here, we performed predictive modeling to examine working memory ability (WMA) as a function of normative whole-brain connectivity across psychiatric diseases. We built a quantitative model for letter three-back task performance in healthy participants, using resting state functional magnetic resonance imaging (rs-fMRI). This normative model was applied to independent participants (N = 965) including four psychiatric diagnoses. Individual’s predicted WMA significantly correlated with a measured WMA in both healthy population and schizophrenia. Our predicted effect size estimates on WMA impairment were comparable to previous meta-analysis results. These results suggest a general association between brain connectivity and working memory ability applicable commonly to health and psychiatric diseases.

We respond to a critique of our temporal Independent Components Analysis (ICA) method for separating global noise from global signal in fMRI data that focuses on the signal versus noise classification of several components. While we agree with several of Power's comments, we provide evidence and analysis to rebut his major criticisms and to reassure readers that temporal ICA remains a powerful and promising denoising approach.

Typical fMRI studies have focused on either the mean trend in the blood‐oxygen‐level‐dependent (BOLD) time course or functional connectivity (FC). However, other statistics of the neuroimaging data may contain important information. Despite studies showing links between the variance in the BOLD time series (BV) and age and cognitive performance, a formal framework for testing these effects has not yet been developed. We introduce the variance design general linear model (VDGLM), a novel framework that facilitates the detection of variance effects. We designed the framework for general use in any fMRI study by modeling both mean and variance in BOLD activation as a function of experimental design. The flexibility of this approach allows the VDGLM to (a) simultaneously make inferences about a mean or variance effect while controlling for the other and (b) test for variance effects that could be associated with multiple conditions and/or noise regressors. We demonstrate the use of the VDGLM in a working memory application and show that engagement in a working memory task is associated with whole‐brain decreases in BOLD variance.

This chapter introduces a neuroimaging approach that has become popular among scientists over the last decade and that now is starting to attain clinical uses. The technique is called resting-state functional connectivity (RSFC), and it typically utilizes functional magnetic resonance imaging (fMRI) scans collected from subjects at rest (i.e., doing nothing) in the scanner. This chapter covers how the data are collected, processed, and turned into measures of brain organization. Some preliminary clinical uses of RSFC are presented. The priority of the chapter is not to broadly survey the field but instead to help readers better understand and more critically assess studies using RSFC techniques.

This article advances two parallel lines of argument about resting-state functional magnetic resonance imaging (fMRI) signals, one empirical and one conceptual. The empirical line creates a four-part organization of the text: (1) head motion and respiration commonly cause distinct, major, unwanted influences (artifacts) in fMRI signals; (2) head motion and respiratory changes are, confoundingly, both related to psychological and clinical and biological variables of interest; (3) many fMRI denoising strategies fail to identify and remove one or the other kind of artifact; and (4) unremoved artifact, due to correlations of artifacts with variables of interest, renders studies susceptible to identifying variance of noninterest as variance of interest. Arising from these empirical observations is a conceptual argument: that an event-related approach to task-free scans, targeting common behaviors during scanning, enables fundamental distinctions among the kinds of signals present in the data, information which is vital to understanding the effects of denoising procedures. This event-related perspective permits statements like "Event X is associated with signals A, B, and C, each with particular spatial, temporal, and signal decay properties". Denoising approaches can then be tailored, via performance in known events, to permit or suppress certain kinds of signals based on their desirability.

There is significant interest in using resting-state functional connectivity (RSFC) to predict human behavior. Good behavioral prediction should in theory require RSFC to be sufficiently distinct across participants; if RSFC were the same across participants, then behavioral prediction would obviously be poor. Therefore, we hypothesize that removing common resting-state functional magnetic resonance imaging (rs-fMRI) signals that are shared across participants would improve behavioral prediction. Here, we considered 803 participants from the human connectome project (HCP) with four rs-fMRI runs. We applied the common and orthogonal basis extraction (COBE) technique to decompose each HCP run into two subspaces: a common (group-level) subspace shared across all participants and a subject-specific subspace. We found that the first common COBE component of the first HCP run was localized to the visual cortex and was unique to the run. On the other hand, the second common COBE component of the first HCP run and the first common COBE component of the remaining HCP runs were highly similar and localized to regions within the default network, including the posterior cingulate cortex and precuneus. Overall, this suggests the presence of run-specific (state-specific) effects that were shared across participants. By removing the first and second common COBE components from the first HCP run, and the first common COBE component from the remaining HCP runs, the resulting RSFC improves behavioral prediction by an average of 11.7% across 58 behavioral measures spanning cognition, emotion and personality. Highlights We decomposed rs-fMRI signals into common subspace & individual-specific subspace Common subspace is shared across all Human Connectome Project (HCP) participants Common subspaces are different across runs, suggesting state-specific effects Individual-specific subspaces are unique to individuals Removal of common subspace signals improve behavioral prediction by 11.7%

There is significant interest in the development and application of deep neural networks (DNNs) to neuroimaging data. A growing literature suggests that DNNs outperform their classical counterparts in a variety of neuroimaging applications, yet there are few direct comparisons of relative utility. Here, we compared the performance of three DNN architectures and a classical machine learning algorithm (kernel regression) in predicting individual phenotypes from whole-brain resting-state functional connectivity (RSFC) patterns. One of the DNNs was a generic fully-connected feedforward neural network, while the other two DNNs were recently published approaches specifically designed to exploit the structure of connectome data. By using a combined sample of almost 10,000 participants from the Human Connectome Project (HCP) and UK Biobank, we showed that the three DNNs do not outperform kernel regression across a wide range of behavioral and demographic measures. Furthermore, the generic feedforward neural network exhibited similar performance to the two state-of-the-art connectome-specific DNNs. We conclude with suggestions on future neuroimaging DNN research, including comparisons with stronger baseline algorithms, minimum sample sizes, transparency of hyperparameter tuning and code availability. Critically, we believe that deep learning remains a promising tool for analyzing neuroimaging data. However, researchers should carefully consider whether and how their applications might benefit from DNNs’ advantages over classical alternatives, rather than treat deep learning as a panacea.

There is significant interest in the development and application of deep neural networks (DNNs) to neuroimaging data. A growing literature suggests that DNNs outperform their classical counterparts in a variety of neuroimaging applications, yet there are few direct comparisons of relative utility. Here, we compared the performance of three DNN architectures and a classical machine learning algorithm (kernel regression) in predicting individual phenotypes from whole-brain resting-state functional connectivity (RSFC) patterns. One of the DNNs was a generic fully-connected feedforward neural network, while the other two DNNs were recently published approaches specifically designed to exploit the structure of connectome data. By using a combined sample of almost 10,000 participants from the Human Connectome Project (HCP) and UK Biobank, we showed that the three DNNs and kernel regression achieved similar performance across a wide range of behavioral and demographic measures. Furthermore, the generic feedforward neural network exhibited similar performance to the two state-of-the-art connectome-specific DNNs. When predicting fluid intelligence in the UK Biobank, performance of all algorithms dramatically improved when sample size increased from 100 to 1000 subjects. Improvement was smaller, but still significant, when sample size increased from 1000 to 5000 subjects. Importantly, kernel regression was competitive across all sample sizes. Overall, our study suggests that kernel regression is as effective as DNNs for RSFC-based behavioral prediction, while incurring significantly lower computational costs. Therefore, kernel regression might serve as a useful baseline algorithm for future studies.

The study of resting state functional connectivity (RSFC) using functional MRI scans has rapidly become one of the most promising and widely used techniques for investigations of human brain function in both healthy and psychiatric populations. Two critical recent developments in this field are 1) the increasing use of simultaneous multi-slice fMRI (multiband) acceleration techniques, which dramatically improves the spatial and temporal resolution of fMRI data, and 2) the recognition that participant motion is a critical confound in RSFC studies, which requires careful denoising in order to obtain valid results. However, motion artifact denoising techniques were not developed with the temporal resolution of multiband fMRI in mind, which results in the capture of high-frequency respiration-related motion of participants during scanning. This respiration-related motion appears to negatively impact the performance of existing volume censoring approaches. Using publicly available multiband RSFC data from the Human Connectome Project, we developed a new volume censoring motion correction approach that addresses respiration-related motion separately from other sources of motion, and outperforms one of the most widely used denoising pipelines. We further show that the assumptions underlying some of the most commonly employed metrics for evaluating motion denoising pipelines (testing for significant differences in RSFC correlations between high- and low-motion participants, and so-called QC-FC based methods) may be inappropriate for the evaluation of pipeline performance. Specifically, the number of significant RSFC correlations between high- and low-motion groups is dramatically reduced by exclusion of participants exhibiting substance use or who have a family history of psychiatric or neurological disorder, indicating that individual differences in unmeasured third variables contribute to both higher motion and true differences in RSFC correlations. As a result, we call into question the use of these widely used metrics as objective, quantitative indicators of data quality after motion denoising, which assumes that no true differences in RSFC exist between high- and low-motion groups. Finally, we develop and present an empirical basis for selecting volume censoring thresholds in any multiband RSFC dataset, which are widely used in the field but have had an exclusively heuristic basis prior to this work. These findings thus have three major impacts: first, to present a substantively improved pipeline for motion denoising of multiband RSFC data; second, to raise concerns about a key metric used to evaluate motion denoising for RSFC data more generally; and third, to provide investigators with an empirically-grounded estimate of the optimal volume censoring threshold to employ for any dataset.

The impact of in-scanner motion on functional magnetic resonance imaging (fMRI) data has a notorious reputation in the neuroimaging community. State-of-the-art guidelines advise to scrub out excessively corrupted frames as assessed by a composite framewise displacement (FD) score, to regress out models of nuisance variables, and to include average FD as a covariate in group-level analyses. Here, we studied individual motion time courses at time points typically retained in fMRI analyses. We observed that even in this set of putatively clean time points, motion exhibited a very clear spatio-temporal structure, so that we could distinguish subjects into separate groups of movers with varying characteristics. Then, we showed that this spatio-temporal motion cartography tightly relates to a broad array of anthropometric and cognitive factors. Convergent results were obtained from two different analytical perspectives: univariate assessment of behavioural differences across mover subgroups unraveled defining markers, while subsequent multivariate analysis broadened the range of involved factors and clarified that multiple motion/behaviour modes of covariance overlap in the data. Our results demonstrate that even the smaller episodes of motion typically retained in fMRI analyses carry structured, behaviourally relevant information. They call for further examinations of possible biases in current regression-based motion correction strategies.

The hippocampus displays a complex organization and function that is perturbed in many neuropathologies. Histological work revealed a complex arrangement of subfields along the medial-lateral and the ventral-dorsal dimension, which contrasts with the anterior-posterior functional differentiation. The variety of maps has raised the need for an integrative multimodal view. We applied connectivity-based parcellation to 1) intrinsic connectivity 2) task-based connectivity, and 3) structural covariance, as complementary windows into structural and functional differentiation of the hippocampus. Strikingly, while functional properties (i.e., intrinsic and task-based) revealed similar partitions dominated by an anterior-posterior organization, structural covariance exhibited a hybrid pattern reflecting both functional and cytoarchitectonic subdivision. Capitalizing on the consistency of functional parcellations, we defined robust functional maps at different levels of partitions, which are openly available for the scientific community. Our functional maps demonstrated a head-body and tail partition, subdivided along the anterior-posterior and medial-lateral axis. Behavioral profiling of these fine partitions based on activation data indicated an emotion-cognition gradient along the anterior-posterior axis and additionally suggested a self-world-centric gradient supporting the role of the hippocampus in the construction of abstract representations for spatial navigation and episodic memory.

The hallmark pathology underlying Parkinson disease (PD) is progressive synucleinopathy, beginning in caudal brainstem that later spreads rostrally. However, the primarily subcortical pathology fails to account for the wide spectrum of clinical manifestations in PD. To reconcile these observations, resting-state functional connectivity (FC) can be used to examine dysfunction across distributed brain networks. We measured FC in a large, single-site study of nondemented PD (N = 107; OFF medications) and healthy controls (N = 46) incorporating rigorous quality control measures and comprehensive sampling of cortical, subcortical and cerebellar regions. We employed novel statistical approaches to determine group differences across the entire connectome, at the network-level, and for select brain regions. Group differences respected well-characterized network delineations producing a striking "block-wise" pattern of network-to-network effects. Surprisingly, these results demonstrate that the greatest FC differences involve sensorimotor, thalamic, and cerebellar networks, with notably smaller striatal effects. Split-half replication demonstrates the robustness of these results. Finally, block-wise FC correlations with behavior suggest that FC disruptions may contribute to clinical manifestations in PD. Overall, these results indicate a concerted breakdown of functional network interactions, remote from primary pathophysiology, and suggest that FC deficits in PD are related to emergent network-level phenomena rather than focal pathology.

The habenula (Hb) inhibits dopaminergic reward signaling in response to negative outcomes and has been linked to numerous functional domains relevant to mental health, including reward prediction, motivation, and aversion processing. Despite its important neuroscientific and clinical implications, however, the human Hb remains poorly understood due to its small size and the associated technical hurdles to in vivo functional magnetic resonance imaging (fMRI) investigation. Using high-resolution 3T fMRI data from 68 healthy young adults acquired through the Human Connectome Project, we developed a rigorous approach for mapping the whole-brain resting-state functional connectivity of the human Hb. Our study combined an optimized strategy for defining subject-level connectivity seeds to maximize Hb BOLD sensitivity with high-quality surface-based alignment for robust functional localization and cortical sensitivity. We identified significant positive Hb connectivity with: (i) conserved brainstem targets, including the dopaminergic ventral tegmental area, serotonergic raphe nuclei, and periaqueductal gray; (ii) subcortical structures related to reward and motor function, including the nucleus accumbens, dorsal striatum, pallidum, thalamus, and cerebellum; and (iii) cortical areas associated with the Salience Network and early sensory processing, including the dorsal anterior cingulate, anterior insula, and primary visual and auditory cortices. Hb connectivity was strongly biased towards task-positive brain regions, with weak or negative connectivity observed throughout the task-negative Default Mode Network. Our study provides a detailed characterization of Hb resting-state functional connectivity in healthy young adults, demonstrating both the feasibility and clinical potential of studying the human Hb using high-resolution 3T fMRI.

The brain exhibits substantial diurnal variation in physiology and function but neuroscience studies rarely report or consider the effects of time of day. Here, we examined variation in resting-state fMRI in around 900 subjects scanned between 8am to 10pm on two different days. Multiple studies across animals and humans have demonstrated that the brain’s global signal amplitude (henceforth referred to as “fluctuation”) increases with decreased arousal. Thus, in accord with known circadian variation in arousal, we hypothesised that global signal fluctuation would be lowest in the morning, increase in the mid-afternoon and dip in the early evening. Instead, we observed a cumulative decrease (22% between 9am to 9pm) in global signal fluctuation as the day progressed. To put the magnitude of this decrease in context, we note that task-evoked fMRI responses are typically in the order of 1% to 3%. Respiratory variation also decreased with time of day, although control analyses suggested that this did not account for the reduction in GS fluctuation. Finally, time of day was associated with marked decreases in resting state functional connectivity across the whole brain. The magnitude of decrease was significantly stronger than associations between functional connectivity and behaviour (e.g., fluid intelligence). These findings reveal unexpected effects of time of day on the resting human brain, which challenge the prevailing notion that the brain’s global signal reflects mostly arousal and physiological artefacts. We conclude by discussing potential mechanisms for the observed diurnal variation in resting brain activity and the importance of accounting for time of day in future studies.

Temporal fluctuations in functional Magnetic Resonance Imaging (fMRI) have been profitably used to study brain activity and connectivity for over two decades. Unfortunately, fMRI data also contain structured temporal “noise” from a variety of sources, including subject motion, subject physiology, and the MRI equipment. Recently, methods have been developed to automatically and selectively remove spatially specific structured noise from fMRI data using spatial Independent Components Analysis (ICA) and machine learning classifiers. Spatial ICA is particularly effective at removing spatially specific structured noise from high temporal and spatial resolution fMRI data of the type acquired by the Human Connectome Project and similar studies. However, spatial ICA is mathematically, by design, unable to separate spatially widespread “global” structured noise from fMRI data (e.g., blood flow modulations from subject respiration). No methods currently exist to selectively and completely remove global structured noise while retaining the global signal from neural activity. This has left the field in a quandary—to do or not to do global signal regression—given that both choices have substantial downsides. Here we show that temporal ICA can selectively segregate and remove global structured noise while retaining global neural signal in both task-based and resting state fMRI data. We compare the results before and after temporal ICA cleanup to those from global signal regression and show that temporal ICA cleanup removes the global positive biases caused by global physiological noise without inducing the network-specific negative biases of global signal regression. We believe that temporal ICA cleanup provides a “best of both worlds” solution to the global signal and global noise dilemma and that temporal ICA itself unlocks interesting neurobiological insights from fMRI data.

Simultaneous multi-slice (multiband) accelerated functional magnetic resonance imaging (fMRI) provides dramatically improved temporal and spatial resolution for resting-state functional connectivity (RSFC) studies of the human brain in health and disease. However, multiband acceleration also poses unique challenges for denoising of subject motion induced data artifacts, the presence of which is a major confound in RSFC research that substantively diminishes reliability and reproducibility. We comprehensively evaluated existing and novel approaches to volume censoring-based motion denoising in the Human Connectome Project (HCP) dataset. We show that assumptions underlying common metrics for evaluating motion denoising pipelines, especially those based on quality control-functional connectivity (QC-FC) correlations and differences between high- and low-motion participants, are problematic, and appear to be inappropriate in their current widespread use as indicators of comparative pipeline performance and as targets for investigators to use when tuning pipelines for their own datasets. We further develop two new quantitative metrics that are instead agnostic to QC-FC correlations and other measures that rely upon the null assumption that no true relationships exist between trait measures of subject motion and functional connectivity, and demonstrate their use as benchmarks for comparing volume censoring methods. Finally, we develop and validate quantitative methods for determining dataset-specific optimal volume censoring parameters prior to the final analysis of a dataset, and provide straightforward recommendations and code for all investigators to apply this optimized approach to their own RSFC datasets.

Significance Spontaneous fMRI signals are used to understand human brain organization throughout the life span and in disease states. Spontaneous fMRI signals contain many artifacts, and removing these artifacts is vital to properly studying neurobiological signals. We report successful removal of a major artifact, spatially focal motion artifact, from resting state fMRI signals via multiecho imaging techniques. By removing motion artifact, we isolate a second kind of motion-associated signal, a respiratory signal, that occurs across the entire brain. We illustrate several techniques that remove this respiratory artifact, yielding fMRI data free of motion-related influences. These two kinds of motion-related signals have distinct physical and spatial bases, and each can strongly and differentially influence signal patterns in fMRI data. “Functional connectivity” techniques are commonplace tools for studying brain organization. A critical element of these analyses is to distinguish variance due to neurobiological signals from variance due to nonneurobiological signals. Multiecho fMRI techniques are a promising means for making such distinctions based on signal decay properties. Here, we report that multiecho fMRI techniques enable excellent removal of certain kinds of artifactual variance, namely, spatially focal artifacts due to motion. By removing these artifacts, multiecho techniques reveal frequent, large-amplitude blood oxygen level-dependent (BOLD) signal changes present across all gray matter that are also linked to motion. These whole-brain BOLD signals could reflect widespread neural processes or other processes, such as alterations in blood partial pressure of carbon dioxide (pCO2) due to ventilation changes. By acquiring multiecho data while monitoring breathing, we demonstrate that whole-brain BOLD signals in the resting state are often caused by changes in breathing that co-occur with head motion. These widespread respiratory fMRI signals cannot be isolated from neurobiological signals by multiecho techniques because they occur via the same BOLD mechanism. Respiratory signals must therefore be removed by some other technique to isolate neurobiological covariance in fMRI time series. Several methods for removing global artifacts are demonstrated and compared, and were found to yield fMRI time series essentially free of motion-related influences. These results identify two kinds of motion-associated fMRI variance, with different physical mechanisms and spatial profiles, each of which strongly and differentially influences functional connectivity patterns. Distance-dependent patterns in covariance are nearly entirely attributable to non-BOLD artifacts.

Resting-state functional magnetic resonance imaging (rs-fMRI) offers the opportunity to non-invasively study brain networks within individuals. Previous individual-specific network mappings do not account for intra-subject (within-subject) variability. Therefore, intra-subject variability might be mistaken for inter-subject (between-subject) differences. Here we propose a multi-session hierarchical Bayesian model (MS-HBM) that explicitly differentiates between intra-subject and inter-subject functional connectivity variability, as well as inter-subject network spatial variability. Across three datasets, sensory-motor networks exhibited lower inter-subject, but higher intra-subject variability than association networks. Compared with other approaches, MS-HBM parcellations generalized better to new rs-fMRI and task-fMRI data. MS-HBM parcellations from a single rs-fMRI session were comparable to a recent template-matching algorithm using five sessions. Individual-specific MS-HBM cortical parcellations were highly reproducible, while capturing inter-subject network differences. Inter-subject differences in the spatial arrangement of cortical networks could predict individuals' cognition, personality and emotion, thus highlighting the importance of examining individual-specific network spatial configuration in rs-fMRI studies.

Resting-state functional magnetic resonance imaging (rs-fMRI) offers the opportunity to delineate individual-specific brain networks. A major question is whether individual-specific network topography (i.e., location and spatial arrangement) is behaviorally relevant. Here, we propose a multi-session hierarchical Bayesian model (MS-HBM) for estimating individual-specific cortical networks and investigate whether individual-specific network topography can predict human behavior. The multiple layers of the MS-HBM explicitly differentiate intra-subject (within-subject) from inter-subject (between-subject) network variability. By ignoring intra-subject variability, previous network mappings might confuse intra-subject variability for inter-subject differences. Compared with other approaches, MS-HBM parcellations generalized better to new rs-fMRI and task-fMRI data from the same subjects. More specifically, MS-HBM parcellations estimated from a single rs-fMRI session (10 minutes) showed comparable generalizability as parcellations estimated by two state-of-the-art methods using five sessions (50 minutes). We also showed that behavioral phenotypes across cognition, personality and emotion could be predicted by individual-specific network topography with modest accuracy, comparable to previous reports predicting phenotypes based on connectivity strength. Network topography estimated by MS-HBM was more effective for behavioral prediction than network size, as well as network topography estimated by other parcellation approaches. Thus, similar to connectivity strength, individual-specific network topography might also serve as a fingerprint of human behavior.

Resting-state functional magnetic resonance imaging (rs-fMRI) offers the opportunity to delineate individual-specific brain networks. A major question is whether individual-specific network topography (i.e., location and spatial arrangement) is behaviorally relevant. Here, we propose a multi-session hierarchical Bayesian model (MS-HBM) for estimating individual-specific cortical networks and investigate whether individual-specific network topography can predict human behavior. The multiple layers of the MS-HBM explicitly differentiate intra-subject (within-subject) from inter-subject (between-subject) network variability. By ignoring intra-subject variability, previous network mappings might confuse intra-subject variability for inter-subject differences. Compared with other approaches, MS-HBM parcellations generalized better to new rs-fMRI and task-fMRI data from the same subjects. More specifically, MS-HBM parcellations estimated from a single rs-fMRI session (10 min) showed comparable generalizability as parcellations estimated by 2 state-of-the-art methods using 5 sessions (50 min). We also showed that behavioral phenotypes across cognition, personality, and emotion could be predicted by individual-specific network topography with modest accuracy, comparable to previous reports predicting phenotypes based on connectivity strength. Network topography estimated by MS-HBM was more effective for behavioral prediction than network size, as well as network topography estimated by other parcellation approaches. Thus, similar to connectivity strength, individual-specific network topography might also serve as a fingerprint of human behavior.